Sunday, May 4, 2008

More about event processing application variety

Back in Blog-land after a few days break due to dental surgery - not a very pleasant experience, but what somewhat smoothed by the service given by the clinic - to provide a reflexology during the surgery - here you can see Oded the reflexologist in action (there is some benefit to go to the most expensive clinic in the city...).

Anyway, I am hopefully recovering - and back online. I've read today in the EP Blogland, the posting of Chris Martins from Apama that described some deals done by Apama outside its "early adopter" type of application - algorithmic trading. This is consistent with previous postings in various Blogs that have talked about the variety of applications that can be assisted by event processing technologies. We in IBM have been facing engagements in variety of industries for variety of cases in this area, with variety of reasons and even different business justifications, we are still studying the variance to try to better classify functional and non-functional requirements to classes of applications. There is one benefit of "early adopters" that it serves as a proof of feasibility to all others, and the capital market applications certainly provided this function for event processing. There is one shortcoming in "early adopters" that it may bias the market in a way that is not really consistent with most applications, since people have tendendcy to employ inductive thinking (sometimes induction with N = 1). In our case the bias has been towards the "ultra low latency" - which is not typical to most applications in other areas. Software platforms that wish to be versatile can expect different requirements (or hacking their original software, and test the customer's patient). More on that - later.

No comments: